Separation properties of theta functions (Q1974934)

Let \(X\) be a non-singular connected projective curve over an algebraically closed field \(k\) and denote by \(U^s\) the set of isomorphism classes of stable vector bundles on \(X\) with given degree \(d\) and rank \(r\). We know, after \textit{C. S. Seshadri} and \textit{D. Mumford}, that \(U^s\) has a natural structure of quasi-projective variety, which can be compactified to a variety \(U\), by adding semistable vector bundles to the boundary. From a set-theoretical point of view \(U=R/G\), where \(R\) is a variety under the action of a reductive group \(G\), and the quotient is endowed with a scheme structure by techniques of geometric invariant theory (GIT for short). Without using GIT, \textit{G. Faltings} [J. Algebr. Geom. 2, 507-568 (1993; Zbl 0790.14019)] showed that \(R\) has enough theta functions to produce a \(G\)-invariant morphism \(\theta:R\to{\mathbb P}^N\), whose image is a closed subvariety \(U_\theta\), and \(\theta\) factors through a map \(\pi:U\to U_\theta\). In particular, there is a bijection between \(U\) and the normalization of \(U_\theta\), which induces a structure of projective variety on \(U\). This paper contains a partial answer to the question, raised by Seshadri, about how close is the normalization map \(\pi:U\to U_\theta\) to being an isomorphism. The author proves that the map \(\pi\) is bijective and is an isomorphism over \(U^s\).